Live Blogging Election Night 2012

November 6, 2012

I'm live blogging election night results as they come in and offering analysis on how the evening is going for each candidate.

12.25pm ET/9.45pm PT There are still a huge number of votes to count in many states, so it's too early to get a complete picture of why the election was called much earlier than I predicted. But among the states with 90% or more of the votes counted, it looks as though the winner (regardless of whether it's Obama or Romney) has outperformed the polls.

For example, Romney is currently ahead by 14% in South Carolina, versus a 13% margin predicted by 538, 34% in Oklahoma versus 32%, 20% in North Dakota versus 15%, 42% in Wyoming versus 37%, and 11% in Indiana versus 9% predicted. Meanwhile, Obama is up by 17% in Illinois versus a predicted 20%, 18% in New Jersey versus 12%, and 6% versus 6% in Pennsylvania. So, on average, Mitt Romney is up by 3% in the states that he has won, while the president-elect, Barack Obama is up by 1% on average in the states he has won. (Note that this is based on a small sample... I'll dig deeper tomorrow.)

With that said, it also looks as though the networks called the races considerably earlier than in previous years. CNN was among the last to call many of the races, but was still quicker than the model predicted. That actually continues something of a trend: 2008 was generally called earlier than 2004, although I suspected at the time that that was because the race wasn't quite as close. Maybe there are other factors at play (like better analysis by the networks, and faster counting by the states).

As I say, I'll take a closer look at all of these things tomorrow and over the next few days. For now, much earlier than I was expecting, I'll be signing off the evening.

11.31pm ET/8.31pm PT Barack Obama is re-elected president, according to most news organizations, at about 11.20pm Eastern Time, well ahead of my projection. I'll dig into the numbers in a bit and see where the projection went wrong. There are basically two possibilities: 1. Obama outperformed the polls, or 2. the networks called the races earlier than in previous years.

11.24pm ET/8.24pm PT ABC calls the race for Obama with a call in Ohio. CNN has called it too.

11.17pm ET/8.17pm PT Oregon called for Obama.

11.16pm ET/8.16pm PT Iowa now called for Obama. That took just over an hour to call. Again much quicker than expected (the model said almost six hours, based on a 3.2% margin for Obama). I'll do some analysis of how the margins are comparing to Nate Silver's projections shortly.

11.10pm ET/8.10pm PT No big surprises from 11pm's calls, although a quick call for Washington is a good sign for Obama. However, CNN has now called Minnesota and Wisconsin for Obama and North Carolina for Romney, all ahead of the model's projection. I'll have to double-check, but I think that Obama is now up 6-2 in early calls. Missouri wasn't called by CNN for Romney as of 11.04pm though, and becomes the first state that is behind schedule for a call (although ABC just called it for Romney a moment ago).

10.50pm ET/7.50pm PT Arizona is called for Romney. It was projected for 12.32am by the model, so again we have an early call, this time for Romney. Obama leads 3-2 in early calls.

10.40pm ET/7.40pm PT The initial flood of results is beginning to slow down as the relatively comfortable wins for each candidate have been called. As that happens, the current results called by CNN are falling back into line with my projection. The states that have been called "early" by that measure are now.

For Obama: Michigan (projected for 11.20pm), Pennsylvania (projected for 11.55pm), New Hampshire (projected for 1.28am)
For Romney: North Dakota (projected for 12.14am)

The model had Missouri "scheduled" to be called for Mitt Romney at 11.04pm. If that time comes and goes without a call from CNN, it's fair to say Romney is doing worse than 538's projection.

10.09pm ET/7.00pm PT CNN calls New Hampshire, three hours ahead of the model's prediction. Results continue to trend ahead of expectations, and this is a big one for Obama, of course.

9.42pm ET/6.42pm PT CNN calls Pennsylvania. The model projected a call at 12.01am, so again we're looking at earlier calls than expected.

9.37pm ET/6.37pm PT As of 9.37pm, races have been called by CNN in the following states not predicted in my final projection:

For Obama: New Jersey, Michigan
For Romney: Georgia, South Dakota, North Dakota

Since the split is fairly even, this lends credence to the theory that (a) Nate Silver's prediction is very close to the mark, and (b) the networks are calling races earlier this year than they have in the past.

9.22pm ET/6.22pm PT ABC calls Pennsylvania. Huge call for Obama at this stage of the evening.

9.21pm ET/6.21pm PT Note that Obama is currently only 4% ahead in Michigan, per CNN, which suggests they're being more aggressive in their calls this year.

9.10pm ET/6.10pm PT The 9pm states were all called according to expectations, but the really big call so far is Michigan. My model had that being called at 11.26pm, based on a projected 7.1% margin. This earlier call means either that CNN is calling states more aggressively than expected, or that Obama is outperforming 538's projection. Either way, we might (might!) have an earlier night than expected.

8.55pm ET/5.55pm PT Arkansas was called immediately, so no further upsets. Tennessee has also now been called for Romney, and that happened pretty quickly, so no big problems for either candidate yet this evening.

Things aren't quite turning out as my original projection had them, mainly because CNN is calling the not-so-close states a bit quicker than expected. The true test will come later this evening as Wisconsin, Michigan, Minnesota etc. etc. come into play, along with the really, really close states.

8.11pm ET/5.11pm PT No call for Romney in Tennessee is a bit of a surprise as he was projected to win by 16.4%. Alabama and Mississippi never call right away (I should have noted that on the chart, my apologies). CNN's call for Obama in Connecticut is good news for him -- much like the early call of Indiana for Romney was good news for him.

At this very early stage, 538's projections seem to be in good shape, and in my horse race, Tennessee becomes the first state to go against expectations, putting Obama 1-0 up.

7.58pm ET/4.58pm PT 8pm is the first big moment of the evening. If Missouri is called for Mitt Romney immediately and Connecticut and New Jersey aren't, that might be a sign that Romney is outperforming expectations.

Throwing a wrench in that analysis, Hurricane Sandy hit both Connecticut and New Jersey, so they might call late for other reasons.

But... if Connecticut and New Jersey are called immediately, it's a sign that Obama is performing ahead of expectations. Or maybe that the networks are just being bolder in calling states than in previous years.

OK. We still won't know anything. Well, I can safely say that if DC isn't called right away, Obama's in trouble.

7.39pm ET/4.39pm PT CNN called Indiana at 7.25pm, by my count, quite a bit earlier than the model's prediction of 8.48pm. A sign that Romney is outperforming expectations?

Maybe.

538's projected margin in Indiana was 8.7%, which puts it in the range where there was a small chance it would be called immediately. A call this quickly, however, suggests the final margin might be closer to 10% for Romney. We'll see.

7.26pm ET/4.26pm PT One important note for those of you who watched my presentation earlier in the day: In the model I used in the presentation, states with a margin less than 15% are not called immediately. In real life, the networks will call states with a margin between 10% and 15% about half of the time. So my original projection is a little pessimistic about how quickly states will be called. In tonight's tracking, I'll put states with expected margins between 7.5% and 15% in the "maybe called" columns for each candidate. If those states are called immediately, it doesn't necessarily mean that the candidate is outperforming expectations -- it may just be that the networks are confident enough about an outcome to call the race quickly.

I'll be tracking the calls in each category through the evening, so we can see how candidates are performing compared to expectations.

7.10pm ET/4.10pm PT A small surprise as Indiana is called immediately by some networks. CNN has not called it though, so far as I can tell. (Ironically, I can't get CNN right now... hoping to fix that shortly.)

7pm ET/4pm PT Each hour (or half hour, sometimes) before polls close, I'll post an analysis of which states are likely be to called for which candidates immediately, which ones will be eventually won by a particular candidate but deemed too close to call by the networks, and which ones are certain (according to my model) to be considered too close to call. Once the networks have made their calls (or not), I'll update the chart.

As a guide, if states that should be called for a candidate are called, that's a good sign for them. If those states aren't called, that's a bad sign. The calling of the "maybes" doesn't necessarily signify much (more on that shortly), while any call of a "too close to call" state at the time the polls close will be a big surprise, and suggest that the election has moved significantly towards one candidate or the other.

As the evening progresses, I'll keep track of what calls were made earlier or later than expected, so we can see which candidate is doing better or worse than anticipated. First set of calls coming up...

Bruce Nash bruce.nash@the-numbers.com

Filed under: Analysis